欢迎来到天天文库
浏览记录
ID:44051036
大小:783.38 KB
页数:73页
时间:2019-10-18
《基于数据融合的电梯故障诊断方法分析》由会员上传分享,免费在线阅读,更多相关内容在工程资料-天天文库。
1、AbstractTheelevator'scompositionisverycomplex.Whilebringconveniencetopeople'sdailylife,theelevatorwillinevitablyappearallkindsoffaults.Therearemanycaseshavetakenplacethatpassengerslosttheirlivesbecauseofelevatorfaultsinthepasttwoyears.Frequentelevatorsafetyaccidentsc
2、atchtheelevatorsafetyregulators,industryassociationsandmanufacturer'sdeepthinking.Peoplealsoputforwardhigherrequirementsforthesafetyofusingtheelevato匚Therefore,itisverymeaningfultodotheresearchinhowtodiagnosisthefaultoftheelevatoreffectivelyInthepaper,themethodofelev
3、atorfaultdiagnosisisstudiedbasedonelevatorjerkfault.Firstly,accordingtothefeaturesoftheelevatorjerkfault,aBPneuralnetworkclassifieroffaultdiagnosiswasdesigned,andtheperformanceoftheclassifierwascomparedunderdifferenttrainingfunctions.OntheBPneuralnetworkiseasytofalli
4、ntolocalminimumvaluewhenbetrained,thegeneticalgorithmwasusedtooptimizetheinitialtrainingweightsandthresholdsofBPneuralnetwork.TheperformanceofoptimizedGA-BPneuralnetworkclassifierwastested.Secondly,theexistingfeatureparametersofelevatorjerkfaultwereoptimizedfromthepe
5、rspectiveofoptimizingfeatureparametersofelevatorjerkfault.Themethodofclusteringanalysiswasfirstusedtoconfirmthatredundantfeatureswereexistedinthecurrentfeatureparametersofelevatorjerkfault.Thenthemainfeatureparameterswereconfirmedfromtheperspectiveofnewmessagefeature
6、vectorandthevariationofthefeatureparameters,andtheredundantfeatureswereremoved.Finally,asinglesamplefeaturedatawasusedtoconductadiagnosismayappearwrongjudgmentduetonoiseinterferenceandotherfactorsintheprocessofpracticalapplication.Themethodofusingmultiplesamplingfeat
7、uredatatoconductadiagnosiswasproposedinthispaper.Throughtheprincipalcomponentanalysismethod,aGA-BPclassifierwasdesignedatthecaseofmultiplesamplingfeaturedatatoconductadiagnosisandthenumberofitsinputswassimplified.Inthepaper,alloftheproposedoptimizationmethodswerevali
8、datedbyusingMATLABsoftwarebasedondatasamplesofelevatorjerkfault.Resultsshowthattheselectedoptimizedfeatureparameterscanbeaccurately
此文档下载收益归作者所有